By Ravi Prakash, Director Product Marketing
The ability to forecast capacity is invaluable for decision making in any industry. If you are a city planner in California tasked with identifying when 85% utilized highways in the bay area will become oversubscribed so you may plan road widening or rerouting projects, having the data-driven equivalent of a crystal ball to forecast highway time-to-zero capacity is crucial.
The same level of visibility into infrastructure capacity is crucial to decision making by IT operations tasked with ensuring that application SLAs are met across silos of multi-vendor infrastructure. Application-centric VirtualWisdom introduced the Capacity Forecast analytic to forecast capacity exhaustion across compute, network and storage. Let us walk through a scenario to see how it works.
You see above the VP of Infrastructure dashboard, one of many role-based dashboard templates included in VirtualWisdom. In the panel “ESX hosts by Open Cases” you notice there are critical alarms in red. Upon selecting one of the critical alarms, you realize that it is on a VMware ESX host called syslab-esx04 which is part of the cluster ProdPlatinum. Drilling down into the topology view, you observe an unusually high number of CPU utilization alarms associated with this server.
Fortunately, you have the Capacity Forecast analytic to help you research further.
From the selection of purpose-built Machine Learning (ML) based analytics you run Capacity Forecast with the following options.
The output of the analytic informs you that there are capacity concerns for CPU, memory and some datastores used by the ESX host.
Reviewing the details further you notice that one of the ESX datastores will reach the user defined capacity threshold within 5 days. This delivers on your goal of predicting capacity exhaustion which allows you to be proactive and take a data driven approach to resource optimization.
This analytic can predict capacity exhaustion based on long term (based on say 6 months of data) as well as short term (based on say 1 month of data) behavior. This is relevant in today’s world where you may bring new applications online which were never used before or you have an inflection point where some of your business is moving for the first time into the public cloud. In such cases the ability to predict capacity exhaustion based on short term behavior is especially useful.
The above is just one example of how Machine Learning based analytics in the AIOps platform VirtualWisdom help you manage infrastructure end-to-end from compute in VMs down to storage.
Like what you hear and wish to learn more? Request a demo